Method and system for forecasting reliability of assets

• US 7,509,235 B2
• Filed: 08/31/2005
• Issued: 03/24/2009
• Est. Priority Date: 08/31/2005
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Aggour, Kareem Sherif, Bonissone, Piero Patrone, Varma, Anil
• Primary Examiner(s): Barbee; Manuel L
• Application Number:
  US11/216,940
• Publication Number:
  US 20070061232A1
• Time in Patent Office: 1,301 Days
• Field of Search: 702 33- 35, 702 81- 84, 702/177, 702179-184, 700 79- 81, 714/1, 714/47
• US Class Current: 702/184
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis

Abstract

A method and system of forecasting reliability of an asset is provided. The method includes identifying peer units of the asset by using selected criteria, performing a search for the peer units based upon the selected criteria, and constructing local predictive models using the peer units. The method also includes estimating the future behavior of the asset based upon the local predictive models and dynamically updating the local predictive models to reflect at least one change in the criteria.

Claims

1. A method of forecasting remaining useful life of an asset, comprising:

identifying peer units of the asset from a set of potential peer units by using selected criteria and performing a search for the peer units based upon the selected criteria;

using a first evolutionary algorithm for processing a defined degree of closeness between said peer units and said asset;

constructing local predictive models using the peer units;

estimating the remaining useful life of the asset based upon the local predictive models; and

updating the local predictive models to reflect at least one change in the criteria using a second evolutionary algorithm.

2. The method of claim 1, wherein the selected criteria includes:

attributes;

search parameters; and

similarity function.

3. The method of claim 2, wherein the attributes include structural features and characteristics of the asset.

4. The method of claim 2, wherein the search parameters include a spread and slope of a distribution of attribute values returned from the search.

5. The method of claim 2, wherein the defined degree of closeness includes applying weights to the selected attributes.

6. The method of claim 2, wherein the attributes include at least one of:

operational data relating to the asset;

maintenance data relating to the asset; and

environmental data relating to the asset.

7. The method of claim 1, wherein the first and second evolutionary algorithm is defined using at least one of a:

filter; and

wrapper.

8. The method of claim 1, wherein updating the local predictive models is performed for selected time frames.

9. The method of claim 1, wherein constructing local predictive models further comprises normalizing and aggregating a set of model outputs.

10. The method of claim 1, further comprising defining one or more best assets according to selected metrics, wherein said selected metrics are selected from at least one of a fixed percentage approach and a fixed number approach.

11. The method of claim 1, further comprising evaluating accuracy of said local predictive models over time and updating the local predictive models based upon said accuracy.

12. A system for estimating remaining useful life of an asset, comprising:

a host system;

a storage device in communication with the host system, the storage device storing asset data with utilization information and maintenance data; and

an asset integrity engine executing on the host system, performing;

identifying peer units of the asset from a set of potential peer units by using selected criteria and performing a search of the storage device for the peer units based upon the selected criteria;

using a first evolutionary algorithm for defining a degree of closeness between said peer units and said asset;

constructing local predictive models using the peer units;

estimating the remaining useful life of the asset based upon the local predictive models; and

updating the local predictive models to reflect at least one change in the criteria using a second evolutionary algorithm.

13. The system of claim 12 wherein the selected criteria includes:

attributes;

search parameters; and

similarity function.

14. The system of claim 13, wherein the attributes include structural features and characteristics of the asset.

15. The system of claim 13, wherein the search parameters include a spread and slope of a distribution of attribute values returned from the search.

16. The system of claim 13, wherein the similarity function includes a defined degree of closeness between the asset and potential peer units, the defined degree of closeness including applying weights to the selected attributes for refining computation of the similarity function.

17. The system of claim 13, wherein the attributes include at least one of:

operational data relating to the asset;

maintenance data relating to the asset; and

environmental data relating to the asset.

18. The system of claim 12, wherein the first and second evolutionary algorithm is defined using at least one of a:

filter; and

wrapper.

19. The system of claim 12, wherein updating the local predictive models is performed for selected time frames.

20. The system of claim 12, wherein one or more best assets are determined according to a fixed percentage approach or a fixed number approach.